Gemini 3.8 Flash

NEW

Key Features

The release focuses on long-horizon coding, agentic work, and multi-step reasoning.
Iterative tool calls support checking and completing complex tasks.
Lower effort settings can reduce reasoning-token overhead.
The model may use additional tokens to improve difficult-task results.
The introductory API input rate is $0.75 per million tokens.
The introductory API output rate is $3.75 per million tokens.
It is a specialized variant for vulnerability detection and automated patching.
The Fairwind Program provides access to selected trusted defenders.

The model improves coding and reasoning through further training and more persistent tool-driven work. On difficult requests it may spend additional tokens checking and refining results, while lower effort settings provide a way to control overhead. The related Flash Cyber variant is tailored to security analysis.


Gemini 3.8 Flash is useful for applications involving autonomous development, research, and structured professional work. Hosted access and usage conditions vary by Google product. Flash Cyber is separately restricted to trusted defenders through the Fairwind Program and should not be assumed available through ordinary Flash access.

Get more likes & reach the top of search results by adding this button on your site!

Embed button preview - Light theme
Embed button preview - Dark theme
TurboType Banner

Subscribe to the AI Search Newsletter

Get top updates in AI to your inbox every weekend. It's free!